https://github.com/miyako/llama-cpp

Local inference engine
https://github.com/miyako/llama-cpp

4d-dependency 4d-llm llama-cpp

Last synced: about 1 month ago
JSON representation

Local inference engine

• Host: GitHub
• URL: https://github.com/miyako/llama-cpp
• Owner: miyako
• License: mit
• Created: 2025-10-25T11:53:53.000Z (5 months ago)
• Default Branch: main
• Last Pushed: 2026-02-21T04:10:37.000Z (about 1 month ago)
• Last Synced: 2026-02-21T11:44:23.136Z (about 1 month ago)
• Topics: 4d-dependency, 4d-llm, llama-cpp
• Language: Ruby
• Homepage: https://miyako.github.io/llama-cpp/
• Size: 52.7 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
![version](https://img.shields.io/badge/version-20%2B-E23089)

![platform](https://img.shields.io/static/v1?label=platform&message=mac-intel%20|%20mac-arm%20|%20win-64&color=blue)

[![license](https://img.shields.io/github/license/miyako/llama-cpp)](LICENSE)

![downloads](https://img.shields.io/github/downloads/miyako/llama-cpp/total)

# llama.cpp

Local inference engine

**aknowledgements**: [ggml-org/llama.cpp](https://github.com/ggml-org/llama.cpp)

### Apple Silicon

* set `BUILD_SHARED_LIBS` to `FALSE`

 

### Intel

* set `GGML_CPU` to `FALSE`

* set `CMAKE_OSX_ARCHITECTURES`  to `x86_64`

### Windows

~~set `LLAMA_CURL` to `FALSE`~~

~~c.f. https://github.com/ggml-org/llama.cpp/issues/9937~~

```

cmake -S . -B build -A x64 ^

 -DBUILD_SHARED_LIBS=FALSE ^

 -DCMAKE_TOOLCHAIN_FILE={...\vcpkg\scripts\buildsystems\vcpkg.cmake} ^

 -DLLAMA_BUILD_SERVER=ON

cmake --build build --config Release

```

* open project sith visual studio

* add curl include paths

* add libraries

```

Crypt32.lib

Secur32.lib

Iphlpapi.lib

libcurl.lib

zlib.lib

ws2_32.lib

``` 

* build each target with `MT`

### Windows ARM

ARM NEON and fp16 C-intrinsics not supported by MSVC native compiler. Use `Clang` or `ninja` instead.

```

cmake -B build -G "Visual Studio 17 2022" -A ARM64 -T ClangCL

 -DCMAKE_SYSTEM_PROCESSOR=ARM64

 -DOPENSSL_ROOT_DIR={arm64 openssl dir}

 -DLLAMA_BUILD_TESTS=OFF

 -DLLAMA_BUILD_SERVER=ON

 -DGGML_OPENMP=OFF

 -DGGML_CCACHE=OFF

 -DBUILD_SHARED_LIBS=FALSE

```

```

cmake -B build

 -DCMAKE_SYSTEM_PROCESSOR=ARM64

 -DOPENSSL_ROOT_DIR={arm64 openssl dir}

 -DLLAMA_BUILD_TESTS=OFF

 -DLLAMA_BUILD_SERVER=ON

 -DGGML_OPENMP=OFF

 -DGGML_CCACHE=OFF

 -DBUILD_SHARED_LIBS=FALSE

```

```

cmake --build build --config Release

```